Assimilation of radiance data at JMA:
recent developments and
prospective plans
K. Okamoto*,
H. Owada, T. Ishibashi and T. Egawa
Numerical Prediction Division (NPD)/
Japan Meteorological Agency (JMA)
JMA
ITSC16, Angra dos Reis, Brazil, 7-13 May 2008
NWP models of JMA (Nov.2007~)
JMA
Model
Global Model (GSM)
Meso-scale Model (MSM)
Resolution
H/V(top height)
TL959 (20km)/60(0.1hPa)
*Before Nov.2007: TL319(60km)/40(0.4hPa)
Forecast range
(Initial time)
84h (00,06,18UTC)
216h (12UTC)
15h (00,06,12,18UTC)
33h (03,09,15,21UTC
Target
1~7 day forecast
Aeronautical forecast
Disaster prevention
information
Data Assimilation
(outer/inner loop)
4D-Var
(TL959/T159 or 20km/80km)
4D-Var
(10/20km)
Window Length
6h
6h
Radiance
Assimilation
RTTOV7, VarBC
X
(T/TCWV/rain retrieval)
5km/50 (21.8km)
Satellite data assimilated
JMA
Sounders
„
Radiance from NOAA15~18/AMSU-A,-B,MHS, Aqua/AMSU-A,
Metop/AMSU-A,MHS in GDAS
‡ Including AP-RARS and EARS
„
Temperature retrieval from NOAA15-18 and Metop/ATOVS in MDAS
Microwave Radiometers (MWRs)
Radiance from DMSP/SSMI, TRMM/TMI and Aqua/AMSR-E in GDAS
„ Total Column Water Vapor (TCWV) and Rain Rate retrieval in MDAS
„
Scatterometer
„
Ocean surface wind vector from QuikSCAT/SeaWinds in GDAS & MDAS
GPS-RO (Radio Occultation)
„
Refractivity from CHAMP in GDAS
AMV (Atmospheric Motion Vector) from Aqua+Terra/MODIS in
GDAS
GDAS
GDAS::Global
GlobalData
DataAssimilation
AssimilationSystem
System
MDAS
:
Meso-scale
Data
Assimilation
Geostationary satellites
MDAS : Meso-scale Data AssimilationSystem
System
AMV from MTSAT-1R, Meteosat7,9, GOES11,12 in GDAS & MDAS
„ CSR (clear sky radiance) from MTSAT-1R in GDAS
„
JMA
MW window channel assimilation
Operationally assimilate less
cloud affected radiances of
SSM/I, TMI and AMSR-E in
GDAS since May 2006
Before
BC
After
BC
Before
BC
After
BC
V.Pol only
„ VarBC
„
test1: Improve QC and BC
Stricter cloud screening &
addition of total column
cloud liquid water
predictor in VarBC
„ Small but consistent
positive impact on the
forecast of T850
„
test2: Add SSMIS+test1
19V, 22V, 37V, 92V
„ Neutral impact from SSMIS
window ch
„
Time
Timesequence
sequenceof
ofRMSE
RMSEof
ofday
day11forecast
forecast
test1
test2
AIRS assimilation
(ongoing development)
Assimilate radiances 54ch/324ch over
the sea
Biases
Biasesof
ofAnalysis/firstAnalysis/firstguess
verified
guess verifiedagainst
against
radiosondes,
Aug2007
radiosondes, Aug2007
„ Channel selection using Entropy Reduction
TT[K]
[K]
(SH)
(SH)
(Rodgers 2000)
„ Detect cloud-affected radiances based on
McNally and Watts (2003)
‡
Comparison with CloudSat showed
performance enough for QC (in poster)
„ Bias correction scheme is the same as
ATOVS except predictors in VarBC
‡
Tbobs, 1/cosθ, const
Positive impact on analysis but neutral
or negative on forecast
„ reduces model biases of
temperature and moisture.
Anal (use AIRS)
Anal (no AIRS)
Guess (use AIRS)
Guess (no AIRS)
JMA
Rh
Rh[%]
[%]
(TRP)
(TRP)
AIRS radiance assimilation
JMA
Degrades forecast skills of the lower tropospheric temperature
„ Possibely caused by the conflict of analysis increment with model trend
Time
Time sequence
sequence of
of RMSE
RMSE of
of 11- &
& 5-day
5-day forecasts
forecasts verified
verified against
against the
the initials
initials
RMSE
RMSEZ500
Z500 N.H.
N.H.
RMSE
RMSET850
T850 Trp
Trp
Other topics on Poster 6.9
JMA
Metop/ATOVS impact
AP-RARS & EARS impact
Clear Sky Radiances (CSRs) of WV-ch from
geostationary satellites
„
1 geo vs. 5 geo
AIRS radiance assimilation
„
CloudSat comparison
Plans
End
JMA
Metop/AMSU-A & MHS assimilation
JMA
Assimilate radiances in GDAS
„ Less cloud-affected, less land-surface-dependent radiances
„ Implement a static scan-dependent Bias Correction (BC) and
adaptive air-mass dependent BC
‡
Variational BC (VarBC) applied to air-mass BC
Significantly positive impacts on forecasts, especially,
of the geopotential height
Moreover, great benefits of early and stable data
dissemination
Forecast
ForecastImprovement
ImprovementRate
Ratewrt
wrtRMSE
RMSEfor
for1-9
1-9day
dayforecasts
forecasts
(cntl-test)/cntl
test
:
using
Metop,
cntl
:
no
Metop
(cntl-test)/cntl test : using Metop, cntl : no Metop
Psea
T850
Z500
Statistically significant
Wsp850
Wsp250
Better
Worse
Forecast hours
0
216
72
144
0
216
72
144
0
216
72
144
0
216
72
144
0
216
72
144
EARS impacts on forecasts
JMA
Forecast
ForecastImprovement
ImprovementRate
Rate average
averageover
over6/11-7/6
6/11-7/62007
2007
Psea
T850
Z500
Wsp850
Wsp250
Better
Temperature
Temperature
improvement
improvementrate
rate
FT=24
FT=24(left)
(left)and
and
120
(right)
120 (right)
Worse
Forecast hours
JMA
Geo. Sat. WVch CSR assimilation
5 Geo.Sat. WVch CSRs added
„ Increase humidity in the mid- to upper troposphere, and even
in the lower troposphere in dry areas
„ Make analysis/guess closer to RAOB around 500hPa in RH
and to AMSU-B&MHS channel 3 in TB
„ Positive impact on forecast, especially at day 1 to 3
„ 5 geo.sat. CSR assimilation has greater positive impact than
one (MTSAT-1R)
T850
Z500
1
1 satellite
satellite
Monthly
Monthly averaged
averaged TCWV
TCWV diff.
diff.
by
by WV-CSR
WV-CSR assimilation
assimilation
Time
Time sequence
sequence of
of
TB
TB O-B
O-B STD
STD
5
5 satellites
satellites